Graduate Students Seminar

Speaker 1: Ryan Lafferty

Title

A new method for estimating missing correlations in distributed studies

Abstract

We propose a methodology for a type of statistical meta-analysis problem that commonly occurs in NIH research partnerships. Suppose we have several labs or research centers which have agreed to collaborate with one another, but are nonetheless unwilling to share their hard-won data which may be expensive and difficult to obtain. Each lab is willing only to provide estimated correlations between variables on which it has collected data. Moreover, each lab may not necessarily have collected data on the same variables. Our method seeks to estimate the correlations between each pair of variables studied by any of the labs, using the aggregate statistics provided. We seek to apply this method to a problem that is currently being investigated at the National Cancer Institute involving a distributed partnership of metabolomic studies.

Speaker 2: Mingkai Yu

Title

State and Parameter Estimation from Partial State Observations in Stochastic Reaction Networks

Abstract

We consider chemical reaction networks modeled by a discrete state and continuous in time Markov process for the vector copy number of the species. We describe new Monte Carlo methods for the accurate and efficient estimation of unobserved states and parameters of a stochastic reaction network based on exact partial state observations in continuous time. We first evolution equations for the conditional distribution, then provide particle filter algorithms, and demonstrate our approach with numerical examples.